Recherche avancée

Sur d’autres sites (4485)

• Non-RSA TLS1.2 Packet decryption

  7 février 2017, par Joseph Wahba

  I am trying to decrypt a pcap file. This pcap file contains a capture of an HLS encrypted video stream. The pcap contains TLSv1.2 packets.

  Below are some information from the pcap file

  Server Hello message Cipher Suite :

  TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384.

  EC Diffie-Hellman server Params : pubkey (1)

  The Certificate Status message :

  Signature Hash Algorithm Hash : SHA256

  Signature Hash Algorithm Signature : ECDSA

  Client Key Exchange Message

  EC Diffie-Hellman server Params : pubkey (2)

  I tried to follow this Wireshark SSL decryption tutorial. But it seems that it works only for RSA encryptions.
  I have been researching for a while and found this discussion. I am quoting an extract from this discussion :

  There is an important parameter to mind : decryption of a passively
  recorded session (with a copy of the server private key) works only if
  the key exchange was of type RSA or static DH ; with "DHE" and "ECDHE"
  cipher suites, you won’t be able to decrypt such a session, even with
  knowledge of the server private key. In that case, you will need
  either the negotiated "master secret", or to use the server private
  key to actively intercept the connection

  It’s note worthy that I have the client private key. In my case, the client is FFmpeg video streamer (FFplay). I had a look also on the TLS v1.2 RFC.

  My question :

  Is it possible to do a decryption in this scenario ? If yes, what do I need to have to do so ?

  Is the decryption done using the client’s private key or using the pre_shared_master (i.e. Diffie-Hellman) ?

• How to extract video clip from larger video based on specific time and duration

  26 juillet 2022, par user19019404

  I have video clips that get created. Each video is recorded for 5 minutes and starts at a time e.g. 10:01:20 to 10:06:19 then the next video from 10:06:20 etc. These videos are recorded at a specific frame rate, be it 5 frames or 30 frames (its dependent on the platform making the recording). The net result is NOT a 5 minute video clip but might be a 2 minute video clip (where everyone moves very quickly in the view as a result of the frame rates). I cannot restrict a 1 to 1 recording as these are generated by external systems.

  


  I need to extract specific portions out of the video. For example I need to extract from 10:03:10 to 10:03:35 (25 seconds). This would equate to 10 seconds into the video up to 14 seconds into the video, in those 4 seconds of video, 25 real world seconds are displayed.

  


  My question is do you have any guidance as to how I can calculate that each second of recording actually means 10 or 12 seconds in real life, therefore go to this frame and record to this frame for example.

  


  I have been looking at cv2.CAP_PROP_FPS to get the video frame rates, CAP_PROP_POS_MSEC and CAP_PROP_FRAME_COUNT and believe the answer might lie there, but not sure.

  


  The thinking being if I work out the frame rate of the video, the total frames then I can divide the one by the other to get to how many frames make up real world seconds (this is where I fail). This way I can run from frame to frame as a result.

  


  Thank you

  


• Why is ffmpeg executable so much faster than using libfdk_aac directly ?

  3 février 2016, par lancejabr

  When I encode a wav file to AAC using the ffmpeg executable (using libfdk_aac and setting -threads 1), the process completes several times faster than when I use the AAC encoding example provided by the libfdk_aac library itself. I looked at the code and the central loop in libfdk_aac is almost identical to the aac_encode_frame method in the ffmpeg source.

  Anyone know why the executable would be so much faster ? I’m looking into compilation options but we’re talking like 8x speedup so I don’t think it’s just putting -O3 on there... Does ffmpeg encode multiple buffers at once even when there is only one thread ?